| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology/MRI (UH-B2A209K), University of Michigan Health
System, 1500 E Medical Center Dr., Ann Arbor, MI 48109-0030.
2 Department of Obstetrics and Gynecology, University of Michigan Health System,
Ann Arbor, MI.
3 Department of Biostatistics, School of Public Health, University of Michigan,
Ann Arbor, MI.
Received September 21, 2006;
accepted after revision June 20, 2007.
Address correspondence to H. K. Hussain.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Images of 103 patients who underwent MRI for suspected müllerian duct anomaly were reviewed. Imaging included axial T1-weighted spin-echo (SE) (TR/TE, 500/10) and sagittal, long-uterine-axis, and short-uterine-axis T2-weighted fast SE (5,000/80) sequences. Agreement between original MRI diagnosis and final clinical diagnosis was assessed using the kappa statistic. Two radiologists retrospectively reviewed all cases with inconsistent MRI and clinical diagnoses to identify causes of discrepancy.
RESULTS. There was excellent agreement (
= 0.8) between MRI
and clinical diagnoses of müllerian duct anomalies. For evaluation of the
uterus, there was agreement in 83 of 103 patients, disagreement in 15 of 103,
and agreement could not be determined in five of 103 because of uncertain MRI
diagnoses. The main causes of disagreement were MRI diagnosis of septate uteri
with two cervices clinically diagnosed as didelphic, partial septate uteri
clinically diagnosed as arcuate, and complex anomalies with features of more
than one class. The main difficulties for MRI were the detection of small
uteri or remnants, characterization of cervical dysgenesis and rare anomalies,
overestimation of cervical mucosal folds, characterization of anomalies in the
presence of fibroids, and delineation of vaginal abnormalities.
CONCLUSION. Despite excellent agreement between MRI and clinical diagnoses of müllerian duct anomalies, there are discrepancies and pitfalls resulting mostly from the absence of a precise and integrated classification scheme, unfamiliarity with rare and complex entities, and suboptimal depiction of some structures on MRI.
Keywords: congenital anomalies • MRI • müllerian duct anomalies • pelvic imaging • reproductive medicine • women's imaging
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The purpose of our study was to retrospectively assess agreement between MRI diagnosis and clinical diagnosis of müllerian duct anomalies and identify causes of discrepancy.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
MRI
All MRI studies were performed on a 1.5-T unit (Signa, GE Healthcare;
maximum gradient strength, 25 mT/m; rise time, 600 milliseconds) and a
4-channel torso phased-array coil. A head coil was used for the infants.
Dedicated pelvic MRI was performed in all patients using the following
sequences: axial T1-weighted spin-echo (SE) imaging (TR range/TE range,
400–650/9–12; slice thickness, 6 mm; interslice gap, 0 mm; matrix,
256 [frequency] x 192 [phase]; number of signal averages [NEX], 2; field
of view, 32–36 cm) and T2-weighted fast SE imaging in the sagittal,
long-axis, and short-axis planes of the uterus (TR range/TE range, 4–6
seconds/80–100 milliseconds; echo-train length, 12; slice thickness, 6
mm; interslice gap, 0 mm; matrix, 512 [frequency] x 224 [phase]; NEX, 4;
field of view, 24–32 cm). Two-dimensional T1-weighted spoiled
gradient-recalled echo imaging before and after gadolinium administration was
performed in a single patient for better delineation of the vagina.
Clinical Diagnosis
The final clinical diagnosis was based on findings at history, pelvic
examination, imaging studies, surgery (laparotomy, laparoscopy, and
hysteroscopy), and clinical follow-up. These examinations were performed
before or after the MRI examination and sometimes were performed at other
institutions before the patients were referred to our institution. Detailed
history and pelvic examination findings were available in all 103
patients.
Forty-two of 103 patients underwent the following procedures: combined laparoscopy and hysteroscopy (n = 24), surgical exploration of the vagina and cervix (n = 6), hysterectomy (n = 5), cesarean delivery (n = 3), laparoscopy (n = 2), myomectomy (n = 1), and combined laparoscopy with surgical exploration of the vagina and cervix (n =1).
Other imaging studies in addition to MRI were performed in 74 of the 103 patients: 49 patients had one additional study, 24 patients had two, and one patient had three. Studies included the following: sonography (n = 65), CT (n = 6), hysterosalpingography (n = 23), hysterosonography with saline infusion of the uterus (n = 4), combined cystography and genitography (n = 1), and vaginography (n =1).
Data Analysis
A single author reviewed the prospective MRI report issued for each patient
to determine the MRI diagnosis. These reports had been issued by a group of
abdominal radiologists who were aware of the patient's history and clinical
data at time of interpretation. The MRI diagnosis was classified as uncertain
when the report did not specify a final diagnosis—for example, if the
conclusion of the report stated bicornuate versus septate uterus. The same
author determined the final clinical diagnoses by reviewing these patients'
electronic medical records.
To identify causes of disagreement between MRI diagnosis and clinical diagnosis, determine reasons of uncertain MRI readings, and identify pitfalls of MRI, two authors with 3 and 8 years of experience in pelvic MRI, respectively, reviewed the MR images of all patients with discrepant or uncertain diagnoses to determine, in consensus, causes of discrepancy and to ascertain whether these cases were classified according to established MRI criteria [1–4] (Table 2). These cases were then discussed in a single meeting with the referring clinicians, all with expertise in müllerian duct anomalies.
|
Statistical Analysis
Agreement between the MRI diagnosis and clinical diagnosis of
müllerian duct anomalies was determined using the kappa statistic, which
was applied to the 10 classes of final diagnoses given in
Table 3, in patients with
definitive MRI diagnosis.
|
Further, all diagnostic results were combined into three categories: first, agreement between MRI diagnosis and clinical diagnosis; second, disagreement; and, third, indeterminate due to uncertain MRI diagnosis. Based on these three categories, a chi-square goodness-of-fit test was performed testing the null hypothesis that these categories occurred solely by chance. Point estimates and 95% CIs on the proportion of agreement, disagreement, and indeterminate cases were also calculated.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
=0.8) (95% CI, 0.71–0.89). As a rule of thumb, =1
indicates perfect agreement, 0.8–< 1 is excellent agreement,
0.6–< 0.8 is good agreement, 0.4–< 0.6 is fair agreement,
> 0–< 0.4 is poor agreement, and = 0 is agreement by
chance alone. The chi-square goodness-of-fit test rejected the null hypothesis
(
2 = 105 on 2 df [p < 1e–16]).
Hence, there is strong evidence that agreement did not occur by chance alone.
Agreement between MRI diagnosis and clinical diagnosis was seen in 81%
(n = 83) of patients (95% CI, 0.73–0.88), disagreement in 14%
(n = 15) (0.078–0.21), and agreement could not be determined
due to uncertain MRI diagnosis in 5% of patients (n = 5)
(0.07–0.9). Table 3 lists the MRI diagnoses as determined from the prospective radiology reports and agreement or disagreement with the final clinical diagnosis for all 103 patients. The 15 patients in whom there was disagreement between the diagnoses are detailed in Table 4. In 10 of these 15 patients (Table 4), the MRI diagnosis was verified as incorrect either by surgery (n = 9) or by vaginoscopy and clinical course (n = 1). Retrospective image review revealed that the incorrect MRI diagnosis was due to poor image quality in one case (No. 1), difficulty in detecting very small structures in two cases (Nos. 2 and 3), difficult image interpretation in four cases (Nos. 4, 5, 8, and 9) (Figs. 1A, 1B, 1C and 2A, 2B), and erroneous image interpretation not in accordance with established MRI criteria in three cases (Nos. 6, 7, and 10) (Fig. 3A, 3B, 3C). Three patients (Nos. 2–4) belong to a subgroup of 19 patients who underwent pelvic MRI to evaluate for the presence of a uterus or uterine remnants, and there was agreement between the MRI and the clinical diagnoses in the remaining 16 of 19 patients (84%).
|
|
|
|
|
|
|
|
|
|
The five cases with uncertain MRI diagnosis are detailed in Table 5. Three of these cases (Nos. 17, 19, and 20) (Fig. 5A, 5B, 5C) were confidently classified by the expert reviewers using established imaging criteria. In one of these patients (Fig. 6), a woman with a unicornuate uterus, MRI provided a complete overview of the complex anatomy that was better than the information obtained from hysteroscopy and laparoscopy combined. In the two remaining cases (Nos. 16 and 18), the MRI findings remained equivocal even in retrospect.
|
|
|
|
|
Vagina
Assessment of the vagina on MRI is not usually requested by the referring
gynecologist because it can be adequately examined clinically. In 12 patients,
however, MRI for evaluation of the vagina was requested by the clinical
service because of difficult or inadequate vaginal examinations. A confident
MRI diagnosis was rendered in 11 of the 12 patients
(Table 6) and was correct in
nine of the 11 patients. The two incorrect MRI diagnoses of vaginal stenosis
and hydrocolpos were in two menstruating adolescent patients with cerebral
palsy [10]. In both patients,
diagnosis of vaginal stenosis was inferred from the presence of a vaginal
cavity distended with simple fluid. Sonography and clinical examination with
the patient under anesthesia showed nonobstructive vaginal collections of
urine in both patients, likely secondary to spasticity of the pelvic floor.
Retrospective review confirmed that the nonobstructive nature of the
hydrocolpos could not be determined from the images.
|
The one patient with uncertain MRI diagnosis of the vagina was a 44-year-old woman with a history of uterus didelphys and remote vaginal septum resection who presented with increasing right pelvic pain and persistent vaginal discharge. MRI confirmed the presence of didelphys uterus with obstruction of the right endometrial cavity, but the exact level of obstruction (cervix or high vagina) could not be determined. The patient underwent surgery and was found to have a high oblique vaginal septum. Retrospective image review confirmed the inability to definitely determine the level of obstruction.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
= 0.8) between MRI diagnosis
and clinical diagnosis of uterine anomalies with agreement in diagnoses in 83
of 103 patients and disagreement in 15. There was also agreement in the
diagnoses in nine of 12 patients referred for evaluation of the vagina. The
MRI diagnosis was uncertain in five of 103 patients evaluated for uterine
anomalies and in one of 12 patients evaluated for vaginal anomalies. The main causes of disagreement between the clinical and radiologic diagnoses were the absence of an integrated clinical–radiologic classification scheme whereby the same criteria are used for diagnosis and the lack of specific imaging or clinical diagnostic criteria. Some errors in radiologic diagnoses resulted from reader unfamiliarity with rare anomalies, inconsistently visualized anatomic structures, and the presence of fibroids that distorted the uterus. Other interpretational errors were caused by poor depiction of some structures on MR images due to either suboptimal signal or resolution of the MR images or suboptimal visualization because of small size (e.g., atretic uterus) or physiologic status (e.g., collapsed vagina).
Three cases of septate uteri with two cervices in our study were clinically diagnosed as didelphic uteri and all three patients declined further investigations. The outer fundal contour is used on MRI to distinguish between duplication anomalies regardless of the presence of single or double cervices. However, the presence of two cervices on clinical examination indicates a didelphic uterus. This uncommon entity of septate uterus with two cervices is included in the American Society of Reproductive Medicine (formerly the American Fertility Society) classification system of müllerian anomalies [9] and is documented in the literature [11–14]. Still, it is not well recognized by some clinicians. The clinical implication of missing this diagnosis is significant because patients can benefit from metroplasty with supracervical resection of the septum [15], especially if there are fertility issues, but this option may be denied if a clinical diagnosis of uterus didelphys is made, which does not require treatment. Thus, the identification of two cervices on clinical examination should prompt further imaging to exclude a septate uterus.
The spectrum of müllerian duct anomalies is a continuum rather than distinct entities, and some complex anomalies have features of more than one class [1]. There are no precise clinical or MRI criteria to enable specific categorization; therefore, such anomalies are often characterized differently by various radiologists and clinicians. This problem was encountered in two patients, including a patient with a radiologic diagnosis of unicornuate uterus and a large communicating rudimentary horn and a clinical diagnosis of bicornuate uterus with asymmetric horns at hysteroscopy. The second patient had a radiologic diagnosis of arcuate uterus, but a broad outer fundal contour on clinical examination led to the suggestion of a probable bicornuate uterus. This clinical impression was not confirmed by hysteroscopy or laparoscopy.
Four patients with arcuate uteri were erroneously diagnosed as having a septate uterus with an incomplete septum on MRI and underwent unnecessary hysteroscopy. This error is due to the absence of specific MRI criteria for these entities. Perhaps the best way to distinguish between the two is by assessing the morphology of the free end of the septum, which will have an acute configuration in the partial septate uterus and an obtuse angle with saddlelike thickening of the inner fundal myometrium in the arcuate uterus [9, 16].
The disagreements in clinical and radiologic diagnoses discussed so far had clinical implications; some patients were denied appropriate surgery and others underwent unnecessary surgery. These diagnostic errors were largely due to the absence of an integrated clinical–radiologic classification scheme and specific diagnostic criteria. These cases highlight the need for communication between the interpreting radiologist and the referring gynecologist.
Difficult interpretation issues were encountered in patients with rare anomalies, such as cervical agenesis, which is a manifestation of the Mayer-Rokitansky-Küster-Hauser (MRKH) syndrome. Although the studies of Reinhold et al. [5] and Lang et al. [17] showed excellent agreement between MRI and surgical findings in patients with MRKH syndrome, both cases of cervical agenesis or dysgenesis in our series were incorrectly assessed, probably due to reader unfamiliarity with this rare anomaly. The incidence of MRKH syndrome among neonate females has been estimated to be between 1:4,000 [18, 19] and 1:10,000 [20], and cases associated with cervical agenesis or dysgenesis constitute a small percentage of these patients. Given its rarity, even specialized MR radiologists may find it challenging to characterize this anomaly. Accurate determination of the presence or absence of the cervix and endocervical canal is important for surgical planning because the absence of a cervix necessitates hysterectomy in this young patient population; an artificial cervix cannot be created.
The longitudinal fold of the cervical mucosa, also known as the plica palmatae, is not always visualized on MRI but can occasionally be seen as a long and thin low-attenuation structure in the cervix on long-uterine-axis T2-weighted images, resulting in an erroneous assumption of a fibrous septum, as was seen in two of our patients. This confusion can occur if the reader is unfamiliar with this structure and can be avoided by carefully inspecting the axial views of the cervix and by following any septum to its proximal attachment; a longitudinal cervical mucosa terminates in the lower uterus and does not have a proximal attachment. Distinguishing between plica palmatae and uterine septum is important to avoid unnecessary hysteroscopy.
There were several uncertain diagnoses due to issues related to MRI. Poor signal-to-noise ratio (SNR) did not allow identification of a thin uterine septum in one of our patients. Identification of a septum is important because such patients may benefit from surgical excision of the septum, especially if there are fertility issues such as recurrent miscarriages. Poor SNR may relate to body habitus, poor coil placement, or very-high-spatial-resolution imaging. Poor SNR can be addressed by increasing the NEX. Reducing spatial resolution also increases SNR, but this change is not desirable when a small structure is being evaluated.
For the subgroup of 19 patients in our series who underwent pelvic MRI to evaluate for the presence of a uterus or uterine remnants, the degree of agreement (16 of 19 patients [84%]) is less than the excellent agreement (100%) reported by Reinhold et al. [5], but it is higher than the results reported by Economy et al. [8]. The latter group of investigators compared MRI and laparoscopic findings in 14 patients with vaginal agenesis and found that MRI successfully predicted uterine anomalies in six of 14 patients and the lack of uterine structures in one of 14 patients. Our study supports the conclusion of Economy et al. in that small uterine remnants may not be seen on MRI or may be seen only retrospectively. This MRI pitfall is probably due to the small size of the remnant and lack of specific tissue characteristics used to identify the uterus; remnants have uniform low signal intensity on T2-weighted imaging without zonal differentiation. Identification of small structures is further complicated by the limited spatial resolution of the 2D sequences used for pelvic imaging.
The vagina is also difficult to evaluate because of its collapsed physiologic status. The presence and level of vaginal obstruction or stenosis are usually inferred from the distal extent of a vaginal fluid collection, if present. However, not all vaginal fluid collections indicate vaginal obstruction or stenosis and could be due to a nonobstructive cause. A vaginal fluid collection was seen in two patients with cerebral palsy and was thought to be due to vaginal spasticity [10]. This entity has also been described by Lang et al. [17] in a patient with cloacal anomaly.
Hemivaginal obstruction in patients with duplication anomalies and ipsilateral renal agenesis [21–30] can usually be confidently characterized with MRI [2, 31–34]. However, as encountered in one of our patients, it can be difficult to determine whether a high obstruction is caused by an oblique vaginal septum, which can be surgically excised, or is due to cervical obstruction, which is more challenging to address. Thus, the knowledge of the level and cause of vaginal obstruction is important for surgical planning. Another cause of vaginal obstruction is imperforate hymen, which is easily diagnosed on MRI. It appears as a complete obstruction caused by a transverse membrane at the junction of the upper two thirds and lower one third of the vagina. Surgical management of these cases is usually straightforward.
Three-dimensional sonography is an important alternative method for the evaluation of müllerian duct anomalies. Although MRI may provide a better overview of the pelvic anatomy and associated renal anomalies, 3D sonography is a cheaper imaging technique that provides high-spatial-resolution images and can better characterize small anatomic structures. It is no longer operator-dependent and the 3D volume is generated by the automatic sweep of the mechanical transducer [35]. However, the performance of sonography can be influenced by body habitus.
Our study has limitations. First, not all of our patients had surgical proof of the MRI diagnosis. We are aware that clinical diagnosis is not the ideal "gold standard," but surgery is not always possible or needed. Only patients with surgically correctable anomalies and those with high clinical suspicion for such anomalies will undergo surgery. Nevertheless, all patients in our study had correlation with pelvic examination and adequate clinical follow-up, and most patients underwent additional imaging studies or office hysteroscopy. As a result of the lack of surgical proof, we may have underestimated the rate of missed uterine septa and remnants on MRI.
Second, the MRI diagnoses in our study were generated by a group of abdominal imagers who have expertise in cross-sectional imaging but are not necessarily experts in gynecologic imaging. These results may more accurately reflect the experience of abdominal radiologists dealing with these rare anomalies in clinical practice, as compared with the previously reported higher MRI accuracy rates by expert reviewers [2–6].
Our results also reflect the outcomes from a larger patient population with wider spectrum of ages and anomalies, including duplication and agenesis anomalies, as compared with the prior studies. The data presented in this study not only reflect the usefulness of MRI for the evaluation of müllerian duct anomalies in routine clinical practice, but also highlight the problems and pitfalls of MRI and the clinical–surgical diagnosis of these anomalies.
In conclusion, MRI is a very useful tool in the evaluation of patients with suspected müllerian duct anomalies. Discrepancies between MRI and clinical diagnoses of müllerian duct anomalies exist and relate to the absence of an integrated classification scheme and to the lack of precise criteria for clinical and imaging classification. Errors in MRI diagnoses also resulted from reader unfamiliarity with rare and complex entities, suboptimal depiction of certain structures on MRI due to small size, lack of specific tissue characteristics or physiologic status, and distortion of uterine anatomy by fibroids. Image interpretation in patients with rare and complex anomalies may pose a problem even for the subspecialized abdominal radiologist, and communication with the referring gynecologists is essential in these cases. Errors in radiologic and clinical diagnoses may have significant clinical implications.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  T M Chandler, L S Machan, P L Cooperberg, A C Harris, and S D Chang Mullerian duct anomalies: from diagnosis to intervention Br. J. Radiol., December 1, 2009; 82(984): 1034 - 1042. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. A. Steinkeler, C. A. Woodfield, E. Lazarus, and M. M. Hillstrom Female Infertility: A Systematic Approach to Radiologic Imaging and Diagnosis RadioGraphics, September 1, 2009; 29(5): 1353 - 1370. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
